Spectrally sensitive PSF shaping for widefield single-molecule microscopy

Widefield single-molecule microscopy is a powerful technique for studying biomolecular complex formation and intramolecular structural transitions. However, the choice of fluorophores used in multicolor experiments is limited by the demands of detection, which limit the flexibility of multicolor experiments and add complexity for each additional wavelength.

Recent work has used point spread function (PSF) engineering to encode spectral information directly into single-molecule microscopy images, ranging from the introduction of a diffraction grating into the imaging pathway to the use of deep learning to optimize a phase mask generated by a spatial light modulator. These proposals have generally focused on distinguishing among discrete lists of 2-5 fluorophores. High levels of categorization accuracy have been demonstrated, but these algorithms suffer from a lack of generality: the phase mask must be reoptimized for every additional fluorophore or new experimental format.

Here, we present progress toward an engineered PSF that accurately maps the spatial profile of a detected fluorophore to the full spectrum of incident fluorescence while minimizing the spatial footprint. We present the results of simulations to determine the possible dynamic range and wavelength resolution of the optimized PSF, outline the architecture of the deep-learning algorithm, and characterize the robustness of the proposed mask against imperfections in the optical system, as a more robust mask will find wider applications in the single-molecule community.